]> git.apps.os.sepia.ceph.com Git - ceph.git/log
ceph.git
15 years agomsgr: Add CEPH_FEATURE_FLOCK to default features set, remove extraneous data from...
Greg Farnum [Mon, 2 Aug 2010 17:39:40 +0000 (10:39 -0700)]
msgr: Add CEPH_FEATURE_FLOCK to default features set, remove extraneous data from protocol

15 years agoceph_fs: use __le64 instead of __u64 for over-the-wire data
Greg Farnum [Thu, 24 Jun 2010 00:11:42 +0000 (17:11 -0700)]
ceph_fs: use __le64 instead of __u64 for over-the-wire data

15 years agomds: new low-level debug output, and turn down previous debugs
Greg Farnum [Tue, 18 May 2010 18:45:26 +0000 (11:45 -0700)]
mds: new low-level debug output, and turn down previous debugs

15 years agomds: recover filelocks properly on client reconnect
Greg Farnum [Tue, 18 May 2010 16:02:27 +0000 (09:02 -0700)]
mds: recover filelocks properly on client reconnect

15 years agomds: debug output and some comments.
Greg Farnum [Mon, 17 May 2010 22:18:24 +0000 (15:18 -0700)]
mds: debug output and some comments.

15 years agomds: locking. Compare iterator to right list!
Greg Farnum [Mon, 17 May 2010 22:17:28 +0000 (15:17 -0700)]
mds: locking. Compare iterator to right list!

15 years agomds: locking. Fix waiting lock count map deletion
Greg Farnum [Mon, 17 May 2010 22:14:58 +0000 (15:14 -0700)]
mds: locking. Fix waiting lock count map deletion

15 years agomds: drop distributed locks on waiting filelock messages, so they don't block
Greg Farnum [Mon, 17 May 2010 22:14:03 +0000 (15:14 -0700)]
mds: drop distributed locks on waiting filelock messages, so they don't block

15 years agomds: better printing on lock messages
Greg Farnum [Fri, 14 May 2010 23:53:12 +0000 (16:53 -0700)]
mds: better printing on lock messages

15 years agomds: replace accidnetal __u64 with uint64_t
Greg Farnum [Fri, 14 May 2010 20:42:07 +0000 (13:42 -0700)]
mds: replace accidnetal __u64 with uint64_t

15 years agomds: Determine if there are file locks based on flockbl size, not dirty cap bit
Greg Farnum [Thu, 13 May 2010 21:20:17 +0000 (14:20 -0700)]
mds: Determine if there are file locks based on flockbl size, not dirty cap bit

15 years agomds: need to size values read from a bufferlist
Greg Farnum [Thu, 13 May 2010 19:04:43 +0000 (12:04 -0700)]
mds: need to size values read from a bufferlist

15 years agomds: handle file locking in MClientCaps on a reconnect.
Greg Farnum [Wed, 12 May 2010 21:08:33 +0000 (14:08 -0700)]
mds: handle file locking in MClientCaps on a reconnect.

15 years agomds: locking. remove ceph_filelock* lists and replace with iter lists
Greg Farnum [Wed, 12 May 2010 18:08:56 +0000 (11:08 -0700)]
mds: locking. remove ceph_filelock* lists and replace with iter lists
This allows for more efficient deletion of elements.

15 years agomds: Fix ceph_lock_state_t to use uint64_t
Greg Farnum [Tue, 11 May 2010 22:52:29 +0000 (15:52 -0700)]
mds: Fix ceph_lock_state_t to use uint64_t

15 years agomds: locks. Fix possible naughty dereference, and fix existence checks
Greg Farnum [Tue, 11 May 2010 22:27:40 +0000 (15:27 -0700)]
mds: locks. Fix possible naughty dereference, and fix existence checks

15 years agomds: locking. Keep trap of neighbors and coalesce them
Greg Farnum [Tue, 11 May 2010 20:01:32 +0000 (13:01 -0700)]
mds: locking. Keep trap of neighbors and coalesce them

15 years agomds: locking. remove zero-valued client held/waiting lock counts from the map.
Greg Farnum [Tue, 11 May 2010 17:22:14 +0000 (10:22 -0700)]
mds: locking. remove zero-valued client held/waiting lock counts from the map.

In the unlikely event that you have a very large cluster doing a lot of
locking of different files amongst the whole cluster, this will keep
memory use down.

15 years agomds: Move lock print function; add debugging; fix fencepost errors.
Greg Farnum [Mon, 10 May 2010 22:42:49 +0000 (15:42 -0700)]
mds: Move lock print function; add debugging; fix fencepost errors.

15 years agoceph_fs: CEPH_LOCK_UNLOCK should have its own bit, for safety
Greg Farnum [Mon, 10 May 2010 22:41:03 +0000 (15:41 -0700)]
ceph_fs: CEPH_LOCK_UNLOCK should have its own bit, for safety

15 years agomds: use the right size mask for file cap bits
Greg Farnum [Fri, 30 Apr 2010 23:45:29 +0000 (16:45 -0700)]
mds: use the right size mask for file cap bits

15 years agoceph_fs: Add cap bits for advisory file locks
Greg Farnum [Fri, 30 Apr 2010 23:24:30 +0000 (16:24 -0700)]
ceph_fs: Add cap bits for advisory file locks

15 years agomds: Remove unnecessary if check.
Greg Farnum [Fri, 30 Apr 2010 19:50:32 +0000 (12:50 -0700)]
mds: Remove unnecessary if check.

We already know that head_in exists, or we would have entered the previous
if-block and returned from the function.

15 years agomds: Now actually requeues blocked locks whenever locks are removed.
Greg Farnum [Thu, 29 Apr 2010 23:09:27 +0000 (16:09 -0700)]
mds: Now actually requeues blocked locks whenever locks are removed.

15 years agomds: fcntl locks can efficiently remove all locks from one client.
Greg Farnum [Fri, 30 Apr 2010 17:13:20 +0000 (10:13 -0700)]
mds: fcntl locks can efficiently remove all locks from one client.

15 years agomds: fcntl. Keep track of how many locks/waiting locks each client has.
Greg Farnum [Thu, 29 Apr 2010 23:01:57 +0000 (16:01 -0700)]
mds: fcntl. Keep track of how many locks/waiting locks each client has.

15 years agomds: implement look_for_lock, for fcntl F_GETLK-style things
Greg Farnum [Thu, 29 Apr 2010 23:00:10 +0000 (16:00 -0700)]
mds: implement look_for_lock, for fcntl F_GETLK-style things

15 years agomds: ceph_filelock and filelock_change need to be explicitly sized
Greg Farnum [Thu, 29 Apr 2010 20:35:46 +0000 (13:35 -0700)]
mds: ceph_filelock and filelock_change need to be explicitly sized

15 years agomds: cleanup
Sage Weil [Thu, 29 Apr 2010 20:14:34 +0000 (13:14 -0700)]
mds: cleanup

15 years agomds: Implement file lock message handling.
Greg Farnum [Thu, 29 Apr 2010 19:02:40 +0000 (12:02 -0700)]
mds: Implement file lock message handling.

15 years agomds: add distributed lock support for fcntl/flock support
Greg Farnum [Wed, 21 Apr 2010 22:01:29 +0000 (15:01 -0700)]
mds: add distributed lock support for fcntl/flock support

15 years agomds: fcntl. add proper encoding to ceph_filelock and ceph_lock_state_t
Greg Farnum [Thu, 29 Apr 2010 17:49:07 +0000 (10:49 -0700)]
mds: fcntl. add proper encoding to ceph_filelock and ceph_lock_state_t

15 years agomds: fnctl. finish remove_locks so it'll activate whichever locks it can.
Greg Farnum [Tue, 27 Apr 2010 00:33:19 +0000 (17:33 -0700)]
mds: fnctl. finish remove_locks so it'll activate whichever locks it can.

This is lazy but it should work.

15 years agomds: fnctl. add get_waiting_overlaps function; fix get_overlapped locks
Greg Farnum [Tue, 27 Apr 2010 00:10:52 +0000 (17:10 -0700)]
mds: fnctl. add get_waiting_overlaps function; fix get_overlapped locks
so that it won't do bad things if there aren't any.

15 years agomds: fnctl. implement half of remove_lock.
Greg Farnum [Mon, 26 Apr 2010 23:35:31 +0000 (16:35 -0700)]
mds: fnctl. implement half of remove_lock.

Still need to handle all the waiting_locks and set them if proper.

15 years agomds: fnctl. implement adjust_locks method, with helpers
Greg Farnum [Fri, 23 Apr 2010 21:27:59 +0000 (14:27 -0700)]
mds: fnctl. implement adjust_locks method, with helpers

15 years agomds: fcntl. implement add_lock function
Greg Farnum [Fri, 23 Apr 2010 18:51:14 +0000 (11:51 -0700)]
mds: fcntl. implement add_lock function
Still need to write some of the helpers

15 years agomds: fcntl. add skeleton [add, remove]_lock functions.
Greg Farnum [Thu, 22 Apr 2010 18:30:34 +0000 (11:30 -0700)]
mds: fcntl. add skeleton [add, remove]_lock functions.
Implement helper functions to be used later.

15 years agomds: add messenging support for advisory file locking
Greg Farnum [Tue, 20 Apr 2010 21:45:30 +0000 (14:45 -0700)]
mds: add messenging support for advisory file locking

15 years agomsg: rename MClientReply::dir_bl to extra_bl. Fix uses
Greg Farnum [Wed, 28 Apr 2010 23:55:11 +0000 (16:55 -0700)]
msg: rename MClientReply::dir_bl to extra_bl. Fix uses

15 years agomds: conditionally include flock metadata in MClientReconnect
Sage Weil [Tue, 11 May 2010 20:31:19 +0000 (13:31 -0700)]
mds: conditionally include flock metadata in MClientReconnect

mds: move flockbl into each reconnect record

Simpler for the client side.

mds: keep old reconnect struct

mds: put flock blob len in ceph_mds_cap_reconnect struct

encode realms the old way

15 years agomds: conditionally include flock metadata in MClientCaps
Sage Weil [Tue, 11 May 2010 20:31:37 +0000 (13:31 -0700)]
mds: conditionally include flock metadata in MClientCaps

15 years agomds: remove size, mtime, atime from cap reconnect
Sage Weil [Tue, 11 May 2010 20:20:49 +0000 (13:20 -0700)]
mds: remove size, mtime, atime from cap reconnect

15 years agomds: make mislinked dentry during replay workaround tolerate directories, too
Sage Weil [Mon, 2 Aug 2010 16:57:12 +0000 (09:57 -0700)]
mds: make mislinked dentry during replay workaround tolerate directories, too

This should probably all go away when the original bug is fixed!

15 years agoauth: fix cast
Wido den Hollander [Sat, 31 Jul 2010 16:14:21 +0000 (09:14 -0700)]
auth: fix cast

15 years agorgw: Support a loglevel of 0
Wido den Hollander [Thu, 29 Jul 2010 11:56:02 +0000 (13:56 +0200)]
rgw: Support a loglevel of 0

15 years agorgw: fix setting acls on the bucket
Yehuda Sadeh [Fri, 30 Jul 2010 23:11:07 +0000 (16:11 -0700)]
rgw: fix setting acls on the bucket

15 years agoauth: define AES iv in common header
Sage Weil [Fri, 30 Jul 2010 17:24:45 +0000 (10:24 -0700)]
auth: define AES iv in common header

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoadd files to .gitignore
Thomas Mueller [Fri, 30 Jul 2010 06:19:53 +0000 (08:19 +0200)]
add files to .gitignore

add generated files to gitignore.

Signed-off-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomonclient: use default port if not specified via -m
Sage Weil [Thu, 29 Jul 2010 22:49:27 +0000 (15:49 -0700)]
monclient: use default port if not specified via -m

Reported-by: Thomas Mueller <thomas@chaschperli.ch>
Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomon: we support MONNAMES feature
Sage Weil [Thu, 29 Jul 2010 20:32:27 +0000 (13:32 -0700)]
mon: we support MONNAMES feature

15 years agoMerge branch 'unstable' into mon-remove
Sage Weil [Thu, 29 Jul 2010 20:24:08 +0000 (13:24 -0700)]
Merge branch 'unstable' into mon-remove

Conflicts:
src/cmon.cc
src/include/ceph_fs.h
src/mon/MonClient.cc
src/mon/Monitor.cc

15 years agomon: fix signed/unsigned compile warning
Sage Weil [Thu, 29 Jul 2010 18:37:17 +0000 (11:37 -0700)]
mon: fix signed/unsigned compile warning

15 years agodebian: add update_pbuilder.sh script; don't do it during build
Sage Weil [Thu, 29 Jul 2010 18:32:19 +0000 (11:32 -0700)]
debian: add update_pbuilder.sh script; don't do it during build

15 years agobump unstable version to v0.22~rc
Sage Weil [Wed, 28 Jul 2010 22:43:31 +0000 (15:43 -0700)]
bump unstable version to v0.22~rc

15 years agov0.21 v0.21
Sage Weil [Tue, 27 Jul 2010 19:43:16 +0000 (12:43 -0700)]
v0.21

15 years agocclass.in: fix libdir definition
Yehuda Sadeh [Wed, 28 Jul 2010 20:59:00 +0000 (13:59 -0700)]
cclass.in: fix libdir definition

15 years agodebian/ceph.logrotate: enhance postrotate for "service" util
Thomas Mueller [Thu, 29 Jul 2010 15:58:17 +0000 (15:58 +0000)]
debian/ceph.logrotate: enhance postrotate for "service" util

[ The following text is in the "UTF-8" character set. ]
    [ Your display is set for the "iso-8859-1" character set.  ]
    [ Some characters may be displayed incorrectly. ]

invoke-rc.d is debian specific.

run "service ceph reload" if invoke-rc.d fails.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agomon: trim old logm states
Sage Weil [Thu, 29 Jul 2010 16:16:57 +0000 (09:16 -0700)]
mon: trim old logm states

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agodebian: ceph-cfuse: move ceph from Recommends to Suggests
Thomas Mueller [Thu, 29 Jul 2010 13:02:07 +0000 (15:02 +0200)]
debian: ceph-cfuse: move ceph from Recommends to Suggests

on a standalone fuse client ceph serverside is not needed. if it's
"Recommended" the default behaviour is to install it too. Move it to "Suggests"
that it's not installed by default if one installs just ceph-fuse.

from debian policy:
Suggests:
This is used to declare that one package may be more useful with one or more
others. Using this field tells the packaging system and the user that the listed
packages are related to this one and can perhaps enhance its usefulness, but
that installing this one without them is perfectly reasonable.

- Thomas

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph.spec.in: create /var/log/ceph/stat dir
Thomas Mueller [Thu, 29 Jul 2010 15:52:05 +0000 (15:52 +0000)]
ceph.spec.in: create /var/log/ceph/stat dir

the logrotate config expects a var/log/ceph/stat dir - create it.

Signed-off-by: Sage Weil <sage@newdream.net>
15 years agoceph.spec.in: rbdtool -> rbd
Sage Weil [Thu, 29 Jul 2010 03:39:52 +0000 (20:39 -0700)]
ceph.spec.in: rbdtool -> rbd

Reported-by: markn1
15 years agodebian: drop gcc, g++ build deps
Sage Weil [Wed, 28 Jul 2010 23:28:27 +0000 (16:28 -0700)]
debian: drop gcc, g++ build deps

15 years agorelease: drop wildcard
Sage Weil [Wed, 28 Jul 2010 23:06:42 +0000 (16:06 -0700)]
release: drop wildcard

15 years agodebian: remove gcc/g++ build-depends
Sage Weil [Wed, 28 Jul 2010 23:01:49 +0000 (16:01 -0700)]
debian: remove gcc/g++ build-depends

(as per Clint Byrum)

15 years agodebian: add trailing newline
Sage Weil [Wed, 28 Jul 2010 22:58:43 +0000 (15:58 -0700)]
debian: add trailing newline

15 years agodebian: fix debhelper depend version
Sage Weil [Wed, 28 Jul 2010 22:51:24 +0000 (15:51 -0700)]
debian: fix debhelper depend version

dh_lintian was introduced in 6.0.7

15 years agodebian: update to latest standards version
Sage Weil [Wed, 28 Jul 2010 22:48:49 +0000 (15:48 -0700)]
debian: update to latest standards version

15 years agodebian: fix weak-library-dev-dependency
Sage Weil [Wed, 28 Jul 2010 22:47:27 +0000 (15:47 -0700)]
debian: fix weak-library-dev-dependency

E: ceph source: weak-library-dev-dependency librados1-dev on librados1
E: ceph source: weak-library-dev-dependency libceph1-dev on libceph1

15 years agocheck_version: don't globber .git_version if we're not in a git repo
Sage Weil [Wed, 28 Jul 2010 22:41:31 +0000 (15:41 -0700)]
check_version: don't globber .git_version if we're not in a git repo

namely, when you extract the release tarball!

15 years agofilestore: add option to disable use of btrfs clone range ioctl
Sage Weil [Wed, 28 Jul 2010 22:00:54 +0000 (15:00 -0700)]
filestore: add option to disable use of btrfs clone range ioctl

15 years agofilestore: fix fallback clone_range implementation on short files
Sage Weil [Wed, 28 Jul 2010 21:57:11 +0000 (14:57 -0700)]
filestore: fix fallback clone_range implementation on short files

If we read less data from the source, don't loop forever... return -ERANGE
instead.

15 years agorelease: put .git_version in release tarball
Sage Weil [Wed, 28 Jul 2010 20:57:43 +0000 (13:57 -0700)]
release: put .git_version in release tarball

15 years agopgmon: Remove creating_pgs which are from deleted pools.
Greg Farnum [Wed, 28 Jul 2010 20:37:02 +0000 (13:37 -0700)]
pgmon: Remove creating_pgs which are from deleted pools.

15 years agoosd: drop broken (and useless) PG::merge_log assert
Sage Weil [Wed, 28 Jul 2010 19:47:21 +0000 (12:47 -0700)]
osd: drop broken (and useless) PG::merge_log assert

The assert should be olog.backlog, not log.backlog.  But it's the same
condition as the if guard that's 3 lines up, making it pretty useless.
Just drop it.

15 years agomds: handle this log replay error better:
Greg Farnum [Wed, 28 Jul 2010 18:54:40 +0000 (11:54 -0700)]
mds: handle this log replay error better:
Log it to the central log so it gets noticed, and
delete the "wrong inode" instead of just unlinking it, to prevent later crashiness.

Should at least band-aid over #312.

15 years agorgw: set default log level to 20
Yehuda Sadeh [Wed, 28 Jul 2010 18:40:59 +0000 (11:40 -0700)]
rgw: set default log level to 20

15 years agolibrados: quiet down some log messages
Yehuda Sadeh [Wed, 28 Jul 2010 18:35:14 +0000 (11:35 -0700)]
librados: quiet down some log messages

15 years agorgw: Support loglevel settings
Wido den Hollander [Wed, 28 Jul 2010 16:48:33 +0000 (18:48 +0200)]
rgw: Support loglevel settings

The gateway now supports changing the verbosity of the logging, this can be achieved by changing the
RGW_LOG_LEVEL envirioment parameter.

For Apache running systems, this can be done with mod_env: SetEnv RGW_LOG_LEVEL 20

This will change the loglevel to the highest level and produce a lot of usefull output for debugging purposes.

For now, all the messages will go to webservers logfile (For Apache: error_log)

15 years agoqa: consistent snaptest-%d.sh naming
Sage Weil [Tue, 27 Jul 2010 20:22:36 +0000 (13:22 -0700)]
qa: consistent snaptest-%d.sh naming

15 years agomds: fix uninitialized LeaseStat for null lease
Sage Weil [Tue, 27 Jul 2010 19:53:20 +0000 (12:53 -0700)]
mds: fix uninitialized LeaseStat for null lease

./include/encoding.h: In member function void
Locker::issue_client_lease(CDentry*, client_t, ceph::bufferlist&, utime_t,
Session*):
 ./include/encoding.h:80: warning: e.LeaseStat::duration_ms may be used
uninitialized in this function

15 years agoosd: log when we get marked down but aren't
Sage Weil [Tue, 27 Jul 2010 19:18:56 +0000 (12:18 -0700)]
osd: log when we get marked down but aren't

15 years agodebug: no name symlink when explicit --log-file
Sage Weil [Tue, 27 Jul 2010 18:20:02 +0000 (11:20 -0700)]
debug: no name symlink when explicit --log-file

15 years agoclient: some whitespace cleanup
Sage Weil [Tue, 27 Jul 2010 17:05:27 +0000 (10:05 -0700)]
client: some whitespace cleanup

15 years agoqa: add localized version of Thomas Mueller's snaptest-2.sh
Greg Farnum [Tue, 27 Jul 2010 19:13:04 +0000 (12:13 -0700)]
qa: add localized version of Thomas Mueller's snaptest-2.sh

http://github.com/vinzent/ceph-testsuite/blob/master/tests/snaptest-2

15 years agorgw: exit after 5 seconds from SIGUSR1 anyway
Yehuda Sadeh [Tue, 27 Jul 2010 18:18:06 +0000 (11:18 -0700)]
rgw: exit after 5 seconds from SIGUSR1 anyway

15 years agorgw: fix that we return data when partial content is requested
Wido den Hollander [Tue, 27 Jul 2010 07:46:37 +0000 (09:46 +0200)]
rgw: fix that we return data when partial content is requested

15 years agorgw: handle sigusr1, get notified for shutdown requests
Yehuda Sadeh [Mon, 26 Jul 2010 23:32:26 +0000 (16:32 -0700)]
rgw: handle sigusr1, get notified for shutdown requests

15 years agorgw: fix returned error code in If-None-Match
Yehuda Sadeh [Mon, 26 Jul 2010 22:34:31 +0000 (15:34 -0700)]
rgw: fix returned error code in If-None-Match

15 years agorgw: Make sure all times are converted to GMT, so we can compare them correctly
Wido den Hollander [Mon, 26 Jul 2010 20:58:26 +0000 (22:58 +0200)]
rgw: Make sure all times are converted to GMT, so we can compare them correctly

15 years agorgw: make sure error structure is initialized
Yehuda Sadeh [Mon, 26 Jul 2010 22:22:41 +0000 (15:22 -0700)]
rgw: make sure error structure is initialized

15 years agorgw: fix date/time format string
Wido den Hollander [Fri, 23 Jul 2010 22:00:13 +0000 (15:00 -0700)]
rgw: fix date/time format string

15 years agorgw: fix return code for prepare_get_obj() on the fs storage
Yehuda Sadeh [Fri, 23 Jul 2010 21:07:02 +0000 (14:07 -0700)]
rgw: fix return code for prepare_get_obj() on the fs storage

15 years agorgw: fix error status code
Yehuda Sadeh [Fri, 23 Jul 2010 21:01:47 +0000 (14:01 -0700)]
rgw: fix error status code

15 years agoWe should sent back the Last-Modified header in GMT as specified by RFC2616
Wido den Hollander [Fri, 23 Jul 2010 14:40:48 +0000 (16:40 +0200)]
We should sent back the Last-Modified header in GMT as specified by RFC2616

15 years agoWe shoud always return the Accept-Ranges header, with a GET or HEAD request, even...
Wido den Hollander [Fri, 23 Jul 2010 13:39:49 +0000 (15:39 +0200)]
We shoud always return the Accept-Ranges header, with a GET or HEAD request, even when the client is not requesting for partial content.

This is to let the client know we accept requests for partial content.

On HEAD requests we should also return the Content-Length header.

And we should also return the Last-Modified header with GET and HEAD requests, this way browsers and proxy's can cache content.

15 years agorgw: support partial content (206) response
Yehuda Sadeh [Thu, 22 Jul 2010 23:48:38 +0000 (16:48 -0700)]
rgw: support partial content (206) response

Couldn't get this one to work yet, as apache (or probably one of
the apache filters) rewrites the content-length value and sets
it to zero.

15 years agomds: Use get_oldest_snap() (not first) in handle_client_lssnap.
Greg Farnum [Mon, 26 Jul 2010 23:43:16 +0000 (16:43 -0700)]
mds: Use get_oldest_snap() (not first) in handle_client_lssnap.

15 years agorgw: remove unused label 'done' (compiler warning)
Greg Farnum [Fri, 23 Jul 2010 23:29:50 +0000 (16:29 -0700)]
rgw: remove unused label 'done' (compiler warning)

15 years agomsgr: fix direct send to loopback connection
Sage Weil [Mon, 26 Jul 2010 22:56:33 +0000 (15:56 -0700)]
msgr: fix direct send to loopback connection

If we send directly to the local/loopback connection, we need to
queue it for local delivery, not go through the usual send paths.

This fixes lost replies to messages sent to self (e.g., reply to
local anchortable).

Signed-off-by: Sage Weil <sage@newdream.net>